Classic Reset SOC on Nightly Reboot

[smanners]

So here is something very interesting.

Exactly four months ago I had a forum topic under wbjr where I stated that on the 19th/20th of May when my midnite classic did its scheduled restart it reset my SOC to 100% and my NetAH to 0. Well this morning being the 19th/20th of September it did exactly the same thing. Damn near flattened my batteries overnight because it reset my SOC and left AUX 1 on when it should have turned it off. This has to be some sort of bug where maybe there is a memiry buffer filling up then resetting thus losing the last SOC % on restart. I tried back then to generate a support ticket and even emailed Midnite Solar but no response. Will have to give it another go. It says xyz in the manual on how to kill you batteries. Well Midnite Solar add this one to your list.

Cheers

[Westbranch]

I believe I was customer #1 (Ryan told me I was "IT")  for testing the NEW Trouble Ticket system.  Worked without any problems for me, so, let us know if it works for you... a good to know item...

[smanners]

I am running the latest firmware on both of my classics.

[ClassicCrazy]

So here is something very interesting.

Exactly four months ago I had a forum topic under wbjr where I stated that on the 19th/20th of May when my midnite classic did its scheduled restart ......

Cheers

What scheduled restart do you mean -?  The Night Auto Reset  which reboots every day at midnight  ?
What do you have the  Reset Net AH on 100% SOC set at ?
Could it have anything to do with the time set on your Classic ?

Larry

[smanners]

Hi,

The scheduled restart is the auto nightly reset. Normally for the other 119 days it does this it happily restarts and the SOC and NetAH are as it where when it restarted thus maintaining perfect operations. However on the 120th  (maybe it has to do with a full moon or something and it goes a bit feral) night for whatever reason it decides it will reset SOC and NetAH thus causing my issues . Its like it forgets what its previous state of charge is as perhaps it fills a memory buffer, clears it but forgets to save the last state.

Not sure if the time setting would have any effect on it. Its out by a couple of minutes but for the last 119  days its been like this and had no issue.

Cheers

[ClassicCrazy]

Weird - you should try a support ticket again so Midnite can try to duplicate the problem you are having.

Larry

[smanners]

Yup... just logged a ticket.

I'm not using SOC for now, gone to Diversion so I can go by voltage. I would prefer SOC but I'm not going to trust SOC until Midnite come up with something. Batteries are way to expensive to risk a deep discharge to often.

[boB]

I'm looking into this now...

Every 120 days at 90 minute intervals (saving/recalling SOC info) would be 1,920 save/recalls not including midnight (23:59)

Adding 120 more to 1,920 = 2041  which IS suspiciously close to 2048, isn't it ?  2048 is binary 2^11....

Hmmmm......   I do not see anything (yet) relating to this value BUT not done yet !

boB

[smanners]

It does sound suspiciously like a memory buffer issue. Technically all it needs to do in code is a round robin if it keeps such information in a list of limited size rather than clear the buffer entirely. But that is pure speculation I have no idea how it handles such data storage.

[boB]

It does sound suspiciously like a memory buffer issue. Technically all it needs to do in code is a round robin if it keeps such information in a list of limited size rather than clear the buffer entirely. But that is pure speculation I have no idea how it handles such data storage.

Well, it's even simpler than that.  All it does is save the value and recall it again when it re-boots.

No counter involved except for the time between 90 minute saves.  And of course, the time at 23:59:59
when the Classic resets.  So, the SOC ~can~ be off by just less than 90 minutes if it happens
to reset inadvertently.

This has to be an obvious and therefore easy fix.  It's just not obvious to me yet.  We have more
brains laying around here waiting to check out this head scratcher so we'll figure it out. Soon I hope.

boB

[smanners]

Hopefully you come to the conclusion that its just my Classic that is acting fruity every 120 days and not a general bug that happens under a very certain set of circumstances. I know how hard that can be to replicate and troubleshoot I have worked in Information Systems for 25 years and seen my fair share of weird stuff.

If it just happened randomly every now and then it would probably be easier becuase in that case it could just be a memory error creeping in caused by hardware malfunction. The fact that it happened exactly 120 days apart would suggest its not though. Is it statistically possible for the classic to reboot just as its trying to save that SOC value thus causing it to lose or corrupt that information? I dont know if the reset is a full system reset or just a very specific part of the system stopping and restarting like a service in an operating system restarting as opposed to a complete operating system restart.

[boB]

The saving of the SOC data and the reboot are precisely timed so that ~shouldn't~ happen.

I know what you're thinking (I think?) and I could see how it might "walk" over itself every 120 days
but it is timed so that it can't do that.  The code saves the data and THEN reboots.  It does some
other things as well of course.

We'll find it. (legs and fingers crossed)

boB

[smanners]

It was what I was thinking.

Feel free to look at the data in mymidnite on my classic. You can see the jump in data values as it ticks over on the 19th/20th if you have such an ability to look at that data. If not I can send you a screen shot.

[boB]

Yeah, go ahead and send or post a screenshot.  Let's see what it is doing.

Thanks,
boB

[smanners]

Screenshot attached. You can clearly see the SOC dropping steadily then all of a sudden jump back to 100%.

Cheers